Allocating Land to New York’s Conservation Reserve Enhancement Program To Maximize Net Environmental Benefits

Laura Jaroszewski, Gregory Poe and Richard N. Boisvert*

(May 2000) Abstract: A programming model is used to assess the welfare effects of regional and practice specifications contained in New York State’s Draft Conservation Reserve Enhancement Program (CREP) proposal. Net social benefits are nearly 75% lower than options that explicitly account for opportunity costs of production, environmental benefits, and participation response functions. Selected Paper to be presented at the annual meetings of the American Agricultural Economics Association in Tampa, Florida, USA, August 2000. Contact Author: Gregory Poe, Department of Agricultural, Resource, and Managerial Economics, Cornell University, Ithaca, NY, USA. E-mail: GLP2@cornell.edu; Tel: 607-255-4707, FAX - 607-255-6696 Key words: benefit cost analysis, conservation programs, participation, linear programming

Copyright 2000 by Laura Jaroszewski, Gregory Poe and Richard N. Boisvert. All rights reserved. Readers may make verbatim copies of this document for non-commercial purposes by any means, provided that this copyright notice appears on al such copies.

* Laura Jaroszewski is a former graduate student, Gregory Poe is an associate professor and Richard N. Boisvert is a professor, Department of Agricultural, Resource, and Managerial Economics, Cornell University, Ithaca, New York.

Allocating Land to New York’s Conservation Reserve Enhancement Program To Maximize Net Environmental Benefits

Background

Motivated by language in the 1996 Farm Bill that Conservation Programs should

maximize the environmental benefits per dollar expended, the USDA established the

Conservation Reserve Enhancement Program (CREP) to reduce agricultural non-point source

pollution in priority areas where the environmental effects are high. To do so, CREP provides

substantially greater financial benefits to participants than does the Conservation Reserve

Program (CRP), and state and federal governments share program costs. As of June 1999,

eight states had approved CREPs, and a number of other states, including New York, were in

various stages of drafting proposals.

Whereas most states’ CREP target a single area of concern, Chesapeake Bay, for

example, New York, in its draft proposal, is proposing a 20,000-acre, statewide program.

Land is to be allocated to 11 resource areas across the state in proportion to the number of

river miles that fail to meet the standards of the Clean Water Act. Enrolled acreage is also

partitioned across eight approved practices (e.g., trees, riparian strips, etc.).

This paper assesses the net social benefits of a program designed under these

specifications for allocating land by region and practice relative to other options that explicitly

account for opportunity costs of production, environmental benefits, and participation. For

the analysis, we develop a programming model to allocate program acreage across the 11

regions and eight practices to maximize net social benefits subject to budget, institutional, and

participation constraints. While this model is an extension of a standard allocation model, the

explicit consideration of participation constraints is unique. In contrast to other models used

2

in similar research, we allow participation rates to vary by region, depending on incentive

payments, farm income, and type of farm enterprise.

In addition to assisting in an evaluation of the acreage allocation strategy in New

York’s proposal, this analytical framework provides program administrators with a means for

examining other policy scenarios in a systematic fashion. Consistent with previous efforts at

the national level (e.g., Heimlich, 1994; Ribaudo et al., 1994), we find that net benefits and

the optimal land allocation can be improved substantially when variations in benefits, costs,

and participation rates across regions and practices are accounted for in program design.

After reviewing New York’s CREP, we describe the conceptual framework and

model. A discussion of the data is followed by the empirical results, including a discussion of

their sensitivity to underlying assumptions. Finally, we present policy implications.

New York State’s Conservation Reserve Enhancement Program

The New York State CREP, designed to improve water quality in areas that suffer

problems due to nutrients, pesticides, and sediments, intends to enroll 20,000 acres that

border streams and water bodies. The proposal highlights the importance of the agricultural

sector to the state’s economy, and as a source of non-point source pollution. Because of the

state’s topography and the location of productive soils, much of New York’s cropland is near

waterways. The problems relate to agriculture’s contribution to non-point source pollution,

unbuffered areas along water bodies, the needs of wildlife, and pollution prevention

opportunities (NYS Department of Agriculture and Markets, 1998).

Based on the draft proposal, the 20,000 acres available through the New York CREP

are to be allocated across 11 regions, which constitute most of the state, and eight

3

conservation practices (USDA, FSA 1998). Allocations are based on percentages of impaired

water bodies by region, and are also specific to authorized practices (Table 1).

Table 1. Total Acreage Allotment by Region and Practice

Resource Area

CREP Acreage Allotment (acres)

A1: Finger Lakes Region 4,340 A2: Lake Ontario Direct Drainage Basin 2,754 A3: Black River / St. Lawrence River Basins 1,929 A4: Lake Erie Direct Drainage Basin 2,454 A5: Mohawk River Basin 1,647 A6: Upper Hudson River Basin 96 A7: Lower Hudson River 1,931 A8: Lake Champlain River Basin 2,311 A9: Chesapeake Bay / Susquehanna River Basins 1,778 A10: Allegany River Basin 690 A11: Peconic River 70 Totals: 20,000

Conservation Practice

CREP Acreage Allotment (acres)

P1: Riparian Buffers 4,842 P2: Filter Strips 9,053 P3: Grass Waterways 105 P4: Contour Grass Strips 4,211 P5: Establishment of Permanent Introduced or Native Grasses

1,053

P6: Tree Planting 210 P7: Diversions 105 P8: Wetland Restoration 421 Totals: 20,000

The Finger Lakes Region is the third largest by area and has the most farms, cropland,

and pasture. It has the largest land allotment. The Peconic River region is the smallest area

by size, and it has only a token allotment. Although it is home to several endangered species,

the Upper Hudson receives only a token allotment. The largest and most populous region, the

Lower Hudson, has less than 10% of the total allotment, while the least populous area, the

Lake Champlain area, receives more than 10%. Nearly 70 percent of the acreage is supposed

4

to be committed to filter strips and riparian buffers; only 1% is in the most expensive

practices, grass waterways and diversions (Jaroszewski, 2000).

In evaluating this CREP proposal, we must understand the implications of these

decisions regarding the allocation of program acreage. To this end, we develop a model to

allocate the acreage, recognizing that social, private, and governmental benefits and costs

differ by region and conservation practice.

The Land Allocation Decision and Model Structure

Decisions made to enroll land in a conservation program that removes agricultural

land from production implicitly reflects a tradeoff between benefits and costs. From society’s

perspective, benefits are due to improved soil, air, and water quality, and through wildlife and

landscape conservation (Ribaudo (1989a, 1989b), Magleby, et al. (1995), and Feather et al.

(1999)). While government payments add to farm income, they are best characterized as

transfer payments that should not be accounted for in benefit cost analyses.

Nevertheless, the benefits come at a cost to taxpayers and lost agricultural production,

which may result in higher commodity prices and a loss of sales in the agribusiness supply

sector. As additional acres are removed from production and placed in conservation

practices, both the program’s benefits and costs increase. The number of acres enrolled in the

program in turn depends on participation rates, which are influenced by the private benefits

and costs to participants.

The private benefits to landowners include the rental value and any governmental cost

share, as well as reduced expenditures for planting, cultivating, and harvesting crops on the

enrolled acres. There might also be productivity increases due to reduced soil erosion, and

5

farmers might value the environmental improvements, but we do not account for these factors

in this analysis. Costs include the value of foregone production and resources needed to

establish and maintain the conservation practice. Such benefits and costs clearly vary by

region and practice (Ribaudo, 1989b).

To reflect these considerations in a model to allocate CREP acreage, let Yij be the

number of acres of land enrolled in the program in region i (i=1,2,…,m) in conservation

practice j (j=1,2,…,n). To maximize net social benefits, we find Yij that:

{ } ∑∑= =

⋅−m

i

n

jijijij

YijYcbMax

1 1

][ , (1)

where bij and cij are the social benefits and costs, respectively, of an acre in region i and

practice j. While many believe that benefits and costs rise non-linearly with acreage enrolled

(rather than linearly as assumed here), our simplifying assumption may provide reasonable

results for allocating this relatively small acreage across the entire state. This allocation

process must also reflect budget and participation constraints and institutional considerations.

Program Limits Constraint: First, land enrolled in all areas and conservation practices,

must not exceed what is available through the program, APL:

∑∑= =

≤m

i

n

j

PLij AY

1 1

. (2)

Budget Constraints: To keep the program within its budget, total government outlays

must only be less than BC, but there also may be legitimate reasons for imposing budget limits

by region and practice. To recognize this possibility, we define wij = budgetary cost to the

government of enrolling an acre from area i in practice j; Wi = total budgetary cost in region

6

i; and Wj = total budgetary cost for practice j. We have separate budget constraints for each

region (3) and practice (4):

0][1

=−∑=

iijij

n

j

WYw , (i=1,2,…,m) and (3)

0][1

=−∑=

m

ijijij WYw , (j=1,2,…,n). (4)

The combined budgets for all practices must also equal the combined budgets committed to

all regions, and these totals must not exceed the overall budget constraint, BC:

∑ ∑= =

≤=n

j

m

iij BCWW

1 1

. (5)

Participation Constraints: Just because public officials offer a land retirement

program, they cannot guarantee widespread participation, and if the incentives offered

through the program are low, participation rates may be low and acreage enrollment goals by

region or practice may not be met. Thus, the farmer’s participation decision, which depends

on a number of factors, including the rental payment and the opportunity cost of participating,

must be in the model. Since the willingness to participate may vary by region and is

unknown, the government must agree initially to enroll acreage over and above its goal for

each region. The final allocation, which satisfies the acreage constraints in equation (2), will

only be known at the end of the enrollment period.

To model this participation behavior, think of Sij as the number of acres offered by the

government in region i for enrollment in practice j. Further, let 1≥ pij ≥ 0 be the probability

that landowners in area i will participate in conservation practice j. Letting Si be the acres

offered in region i and Sj be the acres offered for practice j, we require:

7

0=− ijijij SpY , (i=1,2,…, m; j=1,2,…, n), (6)

∑=

=−n

jiij SS

1

0 , (i=1,2,…,m) and (7)

01

=−∑=

m

ijij SS , (j=1,2,…, n). (8)

From equation (6), we see that only a proportion (equal to the participation rate) of the acres

offered by the government will finally be enrolled in the program by farmers. Equations (7)

and (8) show that the sum of Sij’s over practices equals the total number of acres offered in

each area and that the sum of Sij’s over regions equals the total number of acres offered for

the jth practice. Total acreages allocated to all practices and all regions must be reconciled:

∑ ∑= =

=n

j

m

iij SS

1 1

. (9)

Supply Constraints: Clearly, policymakers cannot contract for more than the eligible land in a

region. Thus, the acres offered through the program to a region, Si , must not exceed the

AiElig, the acres eligible for the program in region i:

Eligii AS ≤ (i=1,2,…,m). (10)

Institutional Considerations: For additional flexibility in conducting policy evaluation,

we also accommodate a number of constraints that reflect various institutional concerns.

Specifically, the idea that decisions might be made to ensure that every region or practice

receives some of the acreage available through the program is addressed.

For example, due to institutional or political considerations, we could set lower and

upper limits (AjLB and Aj

UB, respectively) on the acreage enrolled in practice j:

UBj

m

iij

LBj AYA ∑

=≤≤

1

. (11)

8

These constraints could be to insure certain levels of commitment to a particularly desirable

conservation practice. Similarly, to control for the potential effects on the regions’ agricultural

economies, we could place lower and perhaps, more important, upper limits (AiLB and Ai

UB ,

respectively) on the acreage enrolled in each area i:

UBi

n

jij

LBi AYA ≤≤ ∑

=1

(i=1,2,…,m), (12)

Alternatively, the institutional constraints could be based on some type of weighting

scheme. As is the case in New York’s current CREP proposal, the weights, vi, reflect the ith

region’s contribution to the total number of impaired water bodies:

Yi = vi · APL (i=1,2,…,m), (13)

where APL is the number of acres available through the program from equation (2). The

weights could, of course, be based on other regional characteristics such as the proportion of

farms, farmland, cropland, farm or non-farm population, endangered species, etc.

By solving this model both with and without the constraints in (11) and (12), we

obtain a sense of the cost (in terms of reduced net social benefits) due to institutional or

political considerations inherent in program administrative decisions. To do so, we must have

sufficient data to specify the model empirically. It is to this issue that we now turn.

Data for the Analysis

As with all similar studies, data at the farm and land parcel level ideally suited for the

analysis were not available, but aggregate data for most of the model components are

available by county. Therefore, once the parameters were calculated by county, the data were

9

aggregated for each region, weighting the county estimates by the proportion of acres

harvested (Jaroszewski, 2000). This process disguised some variation within regions, but

continued to reflect substantial differences in resource vulnerabilities and agricultural

production costs and output among regions. Critical data include estimates of social benefits,

governmental costs, social opportunity costs, and participation rates.

Social Benefits: Clearly, estimates of the social, environmental benefits for enrolling

land are among the most important information for this analysis. While there have been

several attempts to estimate the benefits of the CRP nationally,1 there are no similar estimates

specific to New York. To circumvent this problem, we anchor our estimates of

environmental benefits in New York to the most recent national figures, and then adjust them

to reflect differences across the 11 resource areas. Based a benefit-cost analysis required

under Executive Order 12866, environmental benefits from the CRP were estimated at $2.7

billion annually, or $74.18 per acre enrolled (USDA, FSA 1997).2

To begin to capture differences in environmental benefits by region in New York, we

calculated the regional average environmental benefit indices (EBI) reported on acreage

enrolled in CRP over the past several years (USDA, FSA 1997). The USDA has invested

considerable effort in developing the EBI, which has been used by the Agency since 1991 to

rank farmer’s bids for CRP. “Scores are based on the expected environmental improvement

in soil resources, water quality, wildlife habitat, and other resource concerns during the

1 See, for example, Ribaudo (1989b), Ribaudo (1990), Young and Osborn (1990), as well as Magleby, et al. (1995), and Feather, et al. (1999). 2 Benefits are for improvements in soil productivity and water quality and increases in the consumptive and non-consumptive uses of wildlife. The authors stressed that the value is an incomplete measure of benefits; they would most likely be larger if a more comprehensive analysis were performed.

10

time the land is to be enrolled in the program” (USDA, FSA 1999, p.1). 3 To adjust for

the relative differences in benefits, we divided each regional index by the average across

regions. These “normalized” regional-level EBIs were then multiplied by the $74.18 per acre

value (Table 2).

Table 2. Social Benefits Coefficients

Indexed Dollar

Area EBI EBI Benefits

A1 200 1.05 $78.09 A2 183 0.96 $71.58 A3 171 0.90 $66.61 A4 208 1.10 $81.41 A5 225 1.18 $87.85 A6 166 0.87 $64.91 A7 174 0.92 $67.94 A8 153 0.80 $59.62 A9 222 1.17 $86.50 A10 225 1.18 $87.89 A11 163 0.86 $63.89

Mean 190

Budget Coefficients: The budgetary costs, wij, of enrolling an acre include rental, RRi,

and incentive payments, IPi to the landowner, plus the administrative costs of implementing

the program. Since benefits and costs are measured on an annual basis, budget costs include

only the government’s cost share, CSgj, of the annualized portion of establishment costs,

AECij, and normal administrative costs, AC, and a subsidy to help with the costs of

maintaining the practices, MFi, for this program. Hence we have: wij = RRi + IPi + MFi +

CSgj (AECij) + AC. Data on the cost of establishing the practices were available in the USDA’s July 1998

draft proposal for the program (USDA, FSA 1998). Under the current structure of the

policy, the government is only responsible for 50% of these costs. An estimate of $11 per

3 Feather, et al. (1999) used EBI data to allocate CRP acreage in a recent analysis.

11

acre for the administrative costs was also available from the USDA (USDA, FSA 1998). The

proposed incentive payments for CREP effectively double the current rental rate (Table 3).

Table 3. Components of Budget Costs

Area RRI IPI MFi AC A1 $ 37.47 $ 37.47 $ 5.00 $ 11.00 A2 $ 35.39 $ 35.39 $ 5.00 $ 11.00 A3 $ 30.20 $ 30.20 $ 5.00 $ 11.00 A4 $ 35.73 $ 35.73 $ 5.00 $ 11.00 A5 $ 38.29 $ 38.29 $ 5.00 $ 11.00 A6 $ 30.92 $ 30.92 $ 5.00 $ 11.00 A7 $ 39.71 $ 39.71 $ 5.00 $ 11.00 A8 $ 28.48 $ 28.48 $ 5.00 $ 11.00 A9 $ 31.26 $ 31.26 $ 5.00 $ 11.00 A10 $ 33.07 $ 33.07 $ 5.00 $ 11.00 A11 $125.00 $125.00 $ 5.00 $ 11.00

Annualized Establishment Costs by Practice P1 P2 P3 P4 P5 P6 P7 P8

$84.65 $49.48 $596.12 $43.22 $48.43 $31.74 $573.76 $68.55

Social Opportunity Costs: The opportunity cost to society of enrolling an acre

includes foregone income to the farmer and expenditures by the government beyond simple

transfers. The opportunity cost to the government is estimated as part of the budgetary costs.

To account for the opportunity cost to the farmer, it is reasonable to assume for this relatively

small program, there are no increases in agricultural prices due to reduced output. Thus, the

opportunity cost of agricultural production is the value of agricultural output less the variable

costs not incurred if land is enrolled in the CREP.

To summarize, let these opportunity costs per acre be cij, and denote the farmers’

share of establishment costs for practice j as, (1-CSgj ),and NVPi as net value of agricultural

production in region i: cij = {(1- CSgj)· AECj - (RRi + IPi + MFi)} + {wij} + {δ* NVPi},

12

other variables are as above. Since RRi, IPi, and MFi are transfer payments and appear in wij,

as well, we have: cij = AECj + AC + NVPi. Estimated social costs are provided in Table 4.

The term δ*NVPi is the only component that has not been discussed so far. NVPi are

estimates for the harvested acres-weighted averages of revenues per acre less variable costs of

production for seven predominant crops in New York: corn for grain and for silage, alfalfa,

hay other than alfalfa, oats, wheat, and soybeans. Data on yields by county for each crop are

from the New York Agricultural Statistics Service (NYASS 1994-1998). Variable costs of

production are from enterprise budgets (USDA, ERS, 1999c; Pennsylvania State University,

1998; and the Ohio State University, 1999). Prices are converted to 1998 dollars using an

index of prices received by farmers (USDA, ERS 1995, 1998, 1999b). The variable δ, ranging

from 0 to 1, reflects the extent to which acreage placed in the CREP is offset by production

increases elsewhere. If δ=0, the production foregone is offset by increases on the individual

farm or elsewhere, at approximately the same NVP and there is no opportunity cost to

society. In contrast, if δ=1, foregone production is not offset and its value as an opportunity

cost is real. The associated costs associated with each extreme case are presented in Table 4.

However, unless otherwise noted, we assume δ=0 in the analyses that follow.

13

Table 4. Social Opportunity Costs per Acre

All P1 P2 P3 P4 P5 P6 P7 P8 Regions a $95.65 $60.48 $607.12 $54.22 $59.43 $42.74 $584.76 $79.55

a These are social cost coefficients, cij’s, if we ignore the opportunity cost of production.

Region b P1 P2 P3 P4 P5 P6 P7 P8 A1 $223.65 $187.88 $734.52 $181.62 $186.84 $170.15 $712.17 $206.96 A2 $217.37 $182.20 $728.84 $175.94 $181.15 $164.46 $706.48 $201.27 A3 $203.41 $168.24 $714.88 $161.98 $167.19 $150.50 $692.52 $187.31 A4 $227.16 $191.99 $738.63 $185.73 $190.94 $174.25 $716.27 $211.06 A5 $210.82 $175.65 $722.29 $169.36 $174.61 $157.92 $699.94 $194.73 A6 $210.96 $175.79 $722.43 $169.53 $174.74 $158.05 $700.07 $194.86 A7 $199.08 $163.91 $710.55 $157.65 $162.87 $146.17 $688.19 $182.98 A8 $210.52 $175.35 $721.99 $169.09 $174.31 $157.62 $699.64 $194.43 A9 $195.75 $160.58 $707.22 $154.32 $159.54 $142.84 $684.86 $179.65 A10 $217.67 $182.50 $729.14 $176.24 $181.46 $164.77 $706.79 $201.58 A11 $228.70 $193.53 $740.17 $187.27 $192.48 $175.79 $717.81 $212.60

b These are social cost coefficients, cij’s, when the opportunity costs of production are included.

Participation: The inclusion of a participation function in the optimization model is an

important, unique feature of this research, but there was neither the time nor the funding

available to collect original data on which to estimate participation functions specific to the

proposed New York CREP. Following benefits transfer methods used in environmental

valuation, we instead estimated a participation function for New York using data from a

“filter strip” participation survey conducted by Purvis in Nawaygo County, Michigan (Purvis,

1989; Purvis et al., 1989). These data, in contrast to more recent data from research into

conservation program participation (e.g. Lant et al. 1995; Cooper and Keim, 1996; Cooper

and Osborn, 1998; Kingsbury and Boggess, 1999), are more consistent with the New York

situation. The field crops and animal agriculture in Michigan are similar to those in New

York. Purvis also focused on enrolling new lands in filter/buffer strips that were only

expected to constitute a minor proportion of the cropped area in any farm parcel.

14

Purvis used a dichotomous choice contingent participation question to assess which

landowners would enroll in a filter strip program across a range of rental rates from $20 to

$550. Those respondents who answered ‘yes’ to the participation question were asked a

follow-up question concerning the proportion of eligible land that they would actually enroll

at that price. In analyzing these data, Purvis modeled this two stage process in a two-limit

Tobit framework. The dependent variable, bounded by 0 and 1, was the proportion of eligible

acres enrolled in the proposed program. A total of 22 explanatory variables were used,

ranging from objective farm and enterprise data to demographic characteristics and

perceptions of respondents concerning conservation programs.

Despite the similarity between the New York and Michigan situations, it was

impossible to apply Purvis’ equations directly. We had no data for many of the variables and

we needed regional level participation functions Therefore, building on the general

dichotomous choice framework suggested in Hanemann (1984) and the more specific

applications to conservation program participation (Lohr and Park, 1994, 1995), we reframed

the participation decision for use in our model as a discrete choice, random utility framework

in which a subject is assumed to participate in the program if the benefits of participation

exceed the costs.

Assuming a logistic distribution for the error term, a “full” model was estimated using

nine variables for which New York county-level data could be obtained. The explanatory

variables that were retained at this modeling stage include “objective” financial data (e.g., net

payment to farms and household income), and descriptors of the farming enterprise (e.g., total

crop acres and identification of farm enterprises that contributed an important share of the

farm’s cash income). All economic variables were adjusted to 1998 using appropriate indices.

15

As discussed in detail in Jaroszewski (2000) standard likelihood ratio test procedures

and a cutoff significance level of 25% were then used to remove variables with non-significant

coefficients from the analysis. The result was a fairly parsimonious New York Participation

Function, as indicated in Table 5. The coefficients on NETPAY and HAY are positive and

significant. The coefficient on INCOME is also positive, but because of the relatively high

standard error, we are less confident about the size of the effect. As with previous research

on agriculture’s participation in environmental programs, the estimated participation function

is relatively flat across NETPAY values, suggesting that landowners are not highly responsive

to increases and decreases in the incentive payments (see Figure 1). Overall, the mean

willingness to accept compensation is approximately $40, yet the estimated logistic function

suggests that a large proportion of respondents would participate in the program even at a net

financial loss.

Table 5. Logistic New York Participation Function: Pr(yes) = [1 + e -(α +βx)]–1

Variable Definition (variable type)

Estimated Coefficient

(s.e) α Constant -1.4224*

(0.5698) NETPAY Yearly governmental payment for

participating less the private opportunity cost of participating (continuous)

0.0037* (0.0016)

INCOME Annual household income (continuous)

1.230 E-05 (8.69E-06)

HAY Hay is a significant enterprise (=1)

(categorical)

1.1951*

(0.4706)

* indicates significance at the 10% level.

16

Figure 1: The Estimated New York Participation Function

0

0 .1

0 .2

0 .3

0 .4

0 .5

0 .6

0 .7

0 .8

0 .9

-186

-145

-104 -6

3

-22 19 60 101

142

183

224

265

306

347

388

429

470

N ET P A Y ( $ )

Pro

bab

ility

of

Par

tici

pat

ion

To use this estimated function we constructed a NETPAY variable for New York.

Data on rental and other incentive payments, corn for grain prices and yields, as well as the

variable cost of producing corn were needed. Data on the current rental rates and other

incentive fees are reported in Table 3. To estimate the farmers’ opportunity cost in a manner

consistent with Purvis’ analysis, we used five year average corn yields and corn grain prices

(in 1998 dollars as above). County-level data on corn yields from the New York Agricultural

Statistics Service (NYASS 1994, 1995, 1996, 1998) were aggregated to a regional level using

the procedure outlined above. A 1998 estimate for the variable costs of producing corn for

grain in the Northeast is from Pennsylvania State University (1998). These data are

summarized in Table 6. County-level data on the number of acres harvested of hay and the total number of

acres by crop are used from the 1997 Census of Agriculture estimate of the proportion of

cropland in hay. County-level data on median household income in 1989 are available from

the United States Census Bureau's Census of Population and Housing. They are converted

17

to 1998 dollars using the CPI, and aggregated to a regional level as above. A sample of this

data, assuming that CSgj is 100 percent for all practices, is provided in Table 7.

Table 6: Components of private opportunity cost of enrolling land

Area

Price of Corn

($/bushel)

Corn Yield (bushels/acre)

Variable

Cost

Net Returns

A1 2.97 111 $203 $126 A2 2.97 111 $203 $126 A3 2.97 101 $203 $ 96 A4 2.97 112 $203 $130 A5 2.97 105 $203 $108 A6 2.97 108 $203 $119 A7 2.97 104 $203 $106 A8 2.97 109 $203 $121 A9 2.97 102 $203 $100 A10 2.97 105 $203 $110 A11 2.97 120 $203 $153

Table 7: Values for NETPAYij, INCOMEj, HAY and Unadjusted probabilities of

Participation Assuming Cost Share (CSgj) = 100 percent.

NETPAY (CSgj = 100%)

INCOME HAY Participation Rate

(CSgj = 100%)

A1 -$46 $38,545 0.261 0.31 A2 -$50 $38,741 0.256 0.30 A3 -$30 $32,326 0.570 0.39 A4 -$53 $36,535 0.303 0.31 A5 -$27 $33,289 0.494 0.37 A6 -$53 $40,764 0.435 0.35 A7 -$21 $42,734 0.393 0.38 A8 -$59 $35,395 0.444 0.34 A9 -$33 $35,025 0.491 0.37 A10 -$39 $31,400 0.453 0.35 A11 $102 $64,580 0.021 0.44

Based on past participation rates in the CRP, it should be clear that the estimated

participation levels in Table 7 grossly exceed historical participation rates in New York. Two

18

adjustments were made to these rates to more closely reflect the actual situation in the field.

The first adjustment corrects for respondent awareness. Because the survey informed people

about the program's existence and rules, it raised program awareness to 100%. To correct for

this bias, our participation rates were multiplied by an "awareness factor” ranging from 0 to

1. Because CREP is a new program and participation rates have historically been low in the

Northeast for the CRP, we assumed only 25% awareness for our analysis. (Personal

Communication with Allien LaPierre, FSA 1999).

Based on Purvis’ data indicating that people only enrolled a fraction of their awareness

adjusted eligible land, all participation rates were further multiplied by 0.91.

Empirical Analysis and Model Results To conduct the empirical analysis, we established a baseline solution, against which

the sensitivity of the model and policy options in New York’s CREP could be judged. In our

baseline scenario, there are really four separate solutions, or panels (see Table 8). The first

panel (Table 8a) assumes no prior allocation restrictions by region or by practice The other

three panels are solutions assuming regional and practice restrictions indicated in the New

York CREP proposal, both separately and together.

Each of the base solutions is constructed to maximize net social benefits as defined

above, but assuming that social opportunity cost of foregone agricultural production is zero

(i.e., δ=0). To simulate a realistic situation, a budget constraint was set at $2.4 million, the

amount found in New York’s CREP proposal. Because of the limited available acres, it was

necessary to elevate the awareness percentage from 25 to 44% in the Lake Champlain region

and 35% in the Peconic Bay region. These adjustments raise participation rates enough to

ensure a feasible solution to the model in all four baseline situations.

19

In this base scenario, environmental benefits are also assumed not to differ by practice,

and thus it is not surprising that all land is allocated to trees, the least expensive conservation

practice, when no constraints on allocation across practices are imposed (Table 8a). In this

case, land is planted to only five resource areas (Finger Lakes, Lake Erie, Mohawk River,

Chesapeake Bay and Allegany).4 Net social benefits reach nearly $818,000 annually or about

$41 per acre. Annual budgetary costs total nearly $2,024,000 or $101 per acre.

By comparing this solution with the other three in Table 8, we can begin to identify

the implications of the initial guidelines for allocating acreage. If we constrain the allocation

based on the portion of stream miles at risk by region, budgetary costs rise by less than 1%

annually, but net benefits fall by 21% (Table 8c). All land is still planted to trees.

If the only concern from a political standpoint is to ensure that land is distributed by

conservation practice as reflected in the CREP proposal, net benefits would fall by 75%

(Table 8b). The annual budgetary costs would increase by less than $20 per acre. Again, land

would be enrolled in the same regions as above.

If we impose both these sets of a priori allocation rules written into the draft CREP

proposal, budget costs would rise by 16%, and net benefits would plummet by 96% -- from

$41 to just $2 per acre. Clearly, the political desirability of making CREP a statewide

program must be balanced against the increase in net benefits to society by targeting regions

or even smaller areas where the program is most effective.

4 The entries in all tables are the acreage allocated to the region by row and the practice by column. See tables 1 and 2 for names of regions and practices.

Table 8a. Baseline Results With No Institutional Constraints Table 8b. Baseline Results With Constraints By Practice ij 1 2 3 4 5 6 7 8 Totals 1 2 3 4 5 6 7 8 Totals 1 0 0 0 0 0 5,997 0 0 5,997 0 5,997 0 0 0 0 0 0 5,997 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 4 0 0 0 0 0 2,492 0 0 2,492 0 2,071 0 0 0 0 0 421 2,492 5 0 0 0 0 0 2,324 0 0 2,324 0 0 0 2,324 0 0 0 0 2,324 6 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 7 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 8 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 9 0 0 0 0 0 7,430 0 0 7,430 4,842 984 0 1,499 0 0 105 0 7,430 10 0 0 0 0 0 1,756 0 0 1,756 0 0 105 388 1,053 210 0 0 1,756 11 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Total 0 0 0 0 0 20,000 0 0 20,000 4,842 9,053 105 4,211 1,053 210 105 421 20,000

Value Per Acre Value Per Acre % Change Net Benefits: $817,607 $41 Net Benefits: $203,321 $10 -75% Budgetary Costs: $2,023,612 $101 Budgetary Costs: $2,330,755 $117 15%

Table 8c. Baseline Results With Constraints By Region

Table 8d. Baseline Results With Constraints By Region and Practice

ij 1 2 3 4 5 6 7 8 Totals 1 2 3 4 5 6 7 8 Totals 1 0 0 0 0 0 4,340 0 0 4,340 0 4,340 0 0 0 0 0 0 4,340 2 0 0 0 0 0 2,754 0 0 2,754 2,754 0 0 0 0 0 0 0 2,754 3 0 0 0 0 0 1,929 0 0 1,929 402 1,527 0 0 0 0 0 0 1,929 4 0 0 0 0 0 2,454 0 0 2,454 0 1,928 0 0 0 0 105 421 2,454 5 0 0 0 0 0 1,647 0 0 1,647 0 0 0 1,647 0 0 0 0 1,647 6 0 0 0 0 0 96 0 0 96 96 0 0 0 0 0 0 0 96 7 0 0 0 0 0 1,931 0 0 1,931 1,590 0 0 341 0 0 0 0 1,931 8 0 0 0 0 0 2,311 0 0 2,311 0 1,258 0 0 1,053 0 0 0 2,311 9 0 0 0 0 0 1,778 0 0 1,778 0 0 105 1,673 0 0 0 0 1,778 10 0 0 0 0 0 960 0 0 960 0 0 0 480 0 201 0 0 960 11 0 0 0 0 0 70 0 0 70 0 0 0 70 0 0 0 0 70 Total 0 0 0 0 0 20,000 0 0 20,000

4,842 9,053 105 4,211 1,053 210 105 421 20,000

Value Per Acre % Change Value Per Acre % Change Net Benefits: $648,320 $32 -21% Net Benefits: $33,944 $2 -96% Budgetary Costs: $2,040,861 $102 1%

Budgetary Costs: $2,348,004 $117 16%

21

Because of the differential implications of these policy options, it is important to

understand how sensitive the results are to the assumptions in the model. It is to this issue

that we now turn.

Allowing Benefits to Vary By Practice: Our assumption that benefits are constant across

practices is clearly a critical one. To better understand its importance, and based on

conversations with Professor Harold van Es, an environmental scientist at Cornell, we solved

the model again assuming that benefits for riparian buffers and filter strips were scaled up by

25% and values for contour grass strips and trees were scaled down by 25%.

Ignoring the institutional constraints from the CREP proposal, the change in relative

benefits by practice shifts all land enrollment to filter strips. The slightly higher costs of filter

strips compared with trees, are outweighed by the much higher benefits. The allocation of

land across regions is not affected, but net benefits rise to $44 per acre, and budgetary costs

falls slightly to $100 per acre (see Table 9).

To assess how robust the solution is to the adjustment in benefit values by practice,

we performed a series of simulations. According to these simulations, as long as the benefits

to filter strips are less than 20% higher than for trees, the preferred practice remains trees.

However, if the benefits from filter strips are at least 23% greater than trees, the preferred

practice is filter strips. Between 20 and 23%, both practices are employed.

Changing the Assumption About “δ”: Another critical assumption in the base scenario is δ =

0, implying that there is sufficient agricultural land elsewhere in non-environmentally

vulnerable areas and of approximately equal net returns to replace any loss in output from

enrolled acreage. Thus, there is no effect on aggregate supply or the price of agricultural

output and no loss to society from the reduction in agricultural output. This assumption,

Table 9. Results With No Institutional Constraints, Allowing Benefits to Vary Across Practices ij 1 2 3 4 5 6 7 8 Totals 1 0 5,997 0 0 0 0 0 0 5,997 2 0 0 0 0 0 0 0 0 0 3 0 0 0 0 0 0 0 0 0 4 0 2,492 0 0 0 0 0 0 2,492 5 0 2,324 0 0 0 0 0 0 2,324 6 0 0 0 0 0 0 0 0 0 7 0 0 0 0 0 0 0 0 0 8 0 0 0 0 0 0 0 0 0 9 0 7,430 0 0 0 0 0 0 7,430 10 0 1,756 0 0 0 0 0 0 1,756 11 0 0 0 0 0 0 0 0 0 Total 0 20,000 0 0 0 0 0 0 20,000

Total Net Benefits: $881,035; $44 per acre Total Budget Costs: $2,200,957; $110 per acre

Table 10. Results With No Institutional Constraints (δ=1) ij 1 2 3 4 5 6 7 8 Totals 1 0 0 0 0 0 0 0 0 0 2 0 0 0 0 0 0 0 0 0 3 0 0 0 0 0 3,813 0 0 3,813 4 0 0 0 0 0 0 0 0 0 5 0 0 0 0 0 2,324 0 0 2,324 6 0 0 0 0 0 0 0 0 0 7 0 0 0 0 0 4,677 0 0 4,677 8 0 0 0 0 0 0 0 0 0 9 0 0 0 0 0 7,430 0 0 7,430 10 0 0 0 0 0 1,756 0 0 1,756 11 0 0 0 0 0 0 0 0 0 Total 0 0 0 0 0 20,000 0 0 20,000

Total Net Benefits $1,402,234; -$70 per acre Total Budget Costs $1,997,816; $100 per acre

Table 11. Results With No Institutional Constraints: Setting Awareness at 100 Percent ij 1 2 3 4 5 6 7 8 Totals 1 0 0 0 0 0 0 0 0 0 2 0 0 0 0 0 0 0 0 0 3 0 0 0 0 0 0 0 0 3,813 4 0 0 0 0 0 0 0 0 0 5 0 0 0 0 0 9,267 0 0 9,267 6 0 0 0 0 0 0 0 0 0 7 0 0 0 0 0 4,677 0 0 4,677 8 0 0 0 0 0 0 0 0 0 9 0 0 0 0 0 3,678 0 0 3,678 10 0 0 0 0 0 7,024 0 0 7,024 11 0 0 0 0 0 0 0 0 0 Total 0 0 0 0 0 20,000 0 0 20,000

Total Net Benefits $ 897,478; $ 44.87 per acre Total Budget Costs $2,044,020; $102.20 per acre

while perhaps naïve, is either explicitly or

23

implicitly maintained in most analyses conducted of

the CRP.

Alternatively, by assuming that this social opportunity cost is real (δ=1), land allocated

to the Finger Lakes and Lake Erie regions in the baseline model, when no institutional

constraints by region or practice are imposed, moves into the Black/St. Lawrence and Lower

Hudson regions. On average, the opportunity costs of farming are higher in the Finger Lakes

and Lake Erie regions than in the Black/St. Lawrence and Lower Hudson regions (see Table

10).

Perhaps the most significant result from altering these opportunity costs explicitly is

that total net benefits become negative, and would remain so as long as 0.35 ≤ δ ≤ 1. Put

differently, net benefits to society are positive as long as at least 65% of the foregone

agricultural output due to enrollment in the program is accommodated by production

elsewhere. Viewed from a yet a different perspective, as long as we account totally for the

opportunity cost of foregone agricultural output, environmental benefits would have to rise to

$140 per acre for the net social returns of the program to be positive. This is nearly double

the $74 estimate from the CRP used in this study.

Awareness: To explore how the assumption about the level of farmers’ awareness

affects the results, we resolved the base model with no institutional restrictions, but assumed

awareness to be 100%. With full awareness, the value of net benefits increased from $41 per

acre in the baseline panel to $45 per acre, an increase of about 10%. As expected, land

enrolled in the program remains entirely in trees. However, the land was spread across only

three regions, the Mohawk, Chesapeake Bay, and Allegheny, instead of five. No land was

allocated to either the Finger Lakes or the Lake Erie resource areas. (see Table 11)

This result highlights the importance

24

of recognizing the differences in participation and

awareness rates by region. With increased awareness, the supply constraints are not binding

in the Mohawk and Allegheny resource areas and more land is enrolled in regions where net

benefits are higher. Alternatively, when awareness assumption is reduced to 10%, the model

has no feasible solution. Clearly, the additional benefits must be balanced against additional

costs of advertising and promotion to raise awareness.

Summary and Policy Implications

According to mandates in the Conservation Title of the 1996 Farm Bill, geographically

dispersed conservation programs are to be designed to maximize the environmental benefits

per dollar expended. By augmenting the standard allocation model, we provide a framework

for making this type of assessment. Our model not only accounts for differences in social and

private costs and benefits across regions and practices, but also explicitly recognizes how

participation rates differ across regions and are affected by the economic incentives offered by

the program. Using data specific to a proposed CREP in New York, we demonstrate that,

like previous research at the national level, efforts to distribute acreage across the state and

across practices may result in substantial reductions in the net social benefits of the program.

Although the exact numbers reported are specific to our assumptions and the data used, the

broader implications are evident: the political desirability associated with spreading

enrollments across regions and practices must be weighed against potentially large reductions

in realized net benefits.

At a more general policy design scale, the empirical example demonstrates the

importance of recognizing regional differences in participation rates, and identifies a method

for incorporating such considerations into

25

a programming framework. Our results suggest

that these factors could have substantial effects on the enrollment patterns as well as the

economic efficiency of the program.

References:

26

Cooper, Joseph C. and Russ W. Keim. 1996. “Incentive Payments to Encourage Farmer Adoption of Water Quality Protection Practices.” American Journal of Agricultural Economics. 78 (February 1996): 54-64.

Cooper, Joseph C. and C. Tim Osborn. 1998. “The Effect of Rental Rates on the Extension of Conservation Reserve Program Contracts.” American Journal of Agricultural Economics. 80(1): 184-194.

Feather, Peter, Daniel Hellerstein, and LeRoy Hansen. “Economic Valuation of Environmental Benefits and the Targeting of Conservation Programs: The Case of the CRP.” Resource Economics Division, Economic Research Service, U.S. Department of Agriculture. Agricultural Economic Report No. 778. 1999.

Hanemann, Michael W. 1984. “Welfare Evaluations in Contingent Valuation Experiments with Discrete Responses.” American Journal of Agricultural Economics. 66(3): 332-341.

Heimlich, Ralph E. 1994. “Costs of an Agricultural Wetland Reserve.” Land Economics. 70(2): 234-46.

Jaroszewski, Laura Marie, 2000. “Using a Land Allocation Model to Assess the Welfare Economic Implications of New York State’s Conservation Reserve Enhancement Program”, Unpublished Master’s Thesis, Cornell University.

Kingsbury, Leigh and William Boggess. 1999. “An Economic Analysis of Riparian Landowners’ Willingness to Participate in Oregon’s Conservation Reserve Enhancement Program.” A selected paper for the Annual Meeting of the American Agricultural Economics Association. Nashville, Tennessee. August 8-11.

Lant, Christopher L., Steven E. Kraft, and Keith R. Gillman. 1995. “Enrollment of Filter Strips and Recharge Areas in the CRP and USDA Easement Programs.” Journal of Soil and Water Conservation. 50 (March-April): 193-200

Lohr, Luanne and Timothy Park. 1994. “Discrete/Continuous Choices in Contingent Valuation Surveys: Soil Conservation Decisions in Michigan.” Review of Agricultural Economics. 16(1): 1-15.

Lohr, Luanne and Timothy Park. 1995. “Utility Consistent Discrete-Continuous Choices in Soil Conservation.” Land Economics. 71(4): 474-490.

Magleby, Richard, Carmen Sandretto,

27

William Crosswhite, and C. Tim Osborn. “Soil Erosion and Conservation in the United States.” Natural Resources and Environmental Division, Economic Research Service, U.S. Department of Agriculture. Agricultural Information Bulletin No. 718. 1995.

New York Agricultural Statistics Service. 1997 Census of Agriculture. Volume 1. Geographic Area Series, Part 32. AC97-A-32. Issued March 1999.

New York Agricultural Statistics Service, New York State Department of Agricultural and Markets, Division of Statistics in cooperation with U.S. Department of Agriculture, National Agricultural Statistics Service. New York Agricultural Statistics: 1993-1994. Albany. July 1994.

New York Agricultural Statistics Service, New York State Department of Agricultural and Markets, Division of Statistics in cooperation with U.S. Department of Agriculture, National Agricultural Statistics Service. New York Agricultural Statistics: 1994-1995. Albany. July 1995.

New York Agricultural Statistics Service, New York State Department of Agricultural and Markets, Division of Statistics in cooperation with U.S. Department of Agriculture, National Agricultural Statistics Service. New York Agricultural Statistics: 1995-1996. Albany. July 1996.

New York Agricultural Statistics Service, New York State Department of Agricultural and Markets, Division of Statistics in cooperation with U.S. Department of Agriculture, National Agricultural Statistics Service. New York Agricultural Statistics: 1997-1998. Albany. August 1998.

New York State Department of Agriculture and Markets. “Conservation Reserve Enhancement Program (CREP) Proposal for New York State: Proposed Outline.” March 1998.

Ohio State University Cooperative Extension Service. “1999 Ohio Enterprise Crop Budgets.” Online. Available: http://www-agecon.ag.ohio-state.edu/ faculty/rmoore/crops/index.htm. Visited 4 July 1999.

Pennsylvania State University. “The Penn State Agronomy Guide.” University Park: The Pennsylvania State University. 1998. Online. Available: http://agguide. agronomy.psu.edu/sect10/sec102a.htm. Visited 4 July 1999.

Purvis, Amy K. 1989 “An Economic Analysis of Landowners’ Willingness to Enroll Filter Strips in a Conservation Program: A Case Study in Newaygo County, Michigan.” Unpublished M.S. Thesis. Michigan State University.

Purvis, Amy, John P. Hoehn, Vernon L.

28

Sorenson, and Francis J. Pierce. 1989. “Farmers’ Response to a Filter Strip Program: Results for a Contingent Valuation Survey.” Journal of Soil and Water Conservation. 44(5): 501-504.

Ribaudo, Marc O. 1989(a). “Targeting the Conservation Reserve Program to Maximize Water Quality Benefits.” Land Economics. 65(4): 320-332.

Ribaudo, Marc O. 1989(b).“Water Quality Benefits from the Conservation Reserve Program.” Resources and Technology Division, Economic Research Service, U.S. Department of Agriculture. Agricultural Economic Report No. 606. 1989(b).

Ribaudo, Marc O., Daniel Colacicco, Linda L. Langner, Steven Piper, and Glenn D. Schaible. “Natural Resources and Users Benefit from the Conservation Reserve Program.” Resources and Technology Division, Economic Research Service, U.S. Department of Agriculture. Agricultural Economic Report No. 627. 1990.

Ribaudo, Marc O., C. Tim Osborn, and Kazim Konyar. 1994. “Land Retirement as a Tool for Reducing Agricultural Nonpoint Source Pollution.” Land Economics. 70(1): 77-87.

U.S. Department of Agriculture, Economic Research Service. Agricultural Outlook. AO-225. Washington D.C. December 1995

U.S. Department of Agriculture, Economic Research Service. Agricultural Outlook. AGO-257. Washington D.C. December 1998.

U.S. Department of Agriculture, Economic Research Service, Farm Structure and Performance Branch. Updated September 30, 1999(a). “Costs and Returns Reading Room.” Available: http://www.econ.ag.gov/Briefing/FBE/ car/car.htm. Visited 23 June 1999.

U.S. Department of Agriculture, Farm Service Agency. “Conservation Reserve Program--Long-Term Policy: Final Rule.” February 19, 1997(a). Federal Register. Volume 62, Number 33. 7 CFR Parts 704 and 1410. RIN 0560-AE95. pp. 7601-7635.

U.S. Department of Agriculture, Farm Service Agency. “New York State Conservation Reserve Enhancement Program Proposal.” July 1998.

U.S. Department of Agriculture, Economic Research Service, Farm Structure and Performance Branch. Updated September 30, 1999(a). “Costs and Returns Reading Room.” Available: http://www.econ.ag.gov/Briefing/FBE/ car/car.htm. Visited 23 June 1999.

U.S. Department of Agriculture, Farm Service Agency, Conservation and Environmental Programs Division. Updated October 1999(b). “The Conservation Reserve Program.” Available: http://www.fsa.usda.gov/ dafp/cepd/crpinfo.htm.

Young, E. Edwin and C. Tim Osborn.

29

“The Conservation Reserve Program: An Economic Assessment.” Resources and Technology Division, Economic Research Service, U.S. Department of Agriculture. Agricultural Economic Report No. 626. 1990.